User:Xybot/FixStrasseCh

This is the Ruleset FixStrasseCh of the Xybot script

<source lang=perl> sub begin {

 my $me = shift;
 Data::Primitive->setCreator("FixStrasseCh");
 Data::Primitive->setComment('Korrektur der Schreibweise der Strassennamen in der Schweiz; Str(\.?|aße) => Strasse');

}

sub process {

 my ($me, $obj) = @_;

 my $resultstr = "";
 my $work = $obj;
 my $clone = undef;

 die "FATAL  data is outside the bounding box of switzerland - revert any changes already submitted" if (ref($obj) eq "Data::Node" && !$obj->isInBox(45.821, 5.957, 47.809, 10.492));

 foreach $k ("name", "addr:street") {
   my $v=$work->{"tags"}->{$k};
   if (defined($v)) {
     my $gotcha=0;
     if( $v =~ /Ã./) {
       $gotcha++ if ($v =~ s/ä/ä/);									# Potlatch Krankheit ä
       $gotcha++ if ($v =~ s/ö/ö/);									# Potlatch Krankheit ö
       $gotcha++ if ($v =~ s/ü/ü/);									# Potlatch Krankheit ü
       $gotcha++ if ($v =~ s/Ä/Ä/);									# Potlatch Krankheit Ä
       $gotcha++ if ($v =~ s/Ö/Ö/);									# Potlatch Krankheit Ö
       $gotcha++ if ($v =~ s/Ãœ/Ü/);									# Potlatch Krankheit Ü
       $gotcha++ if ($v =~ s/ß/ß/);									# Potlatch Krankheit ß
     }
     $gotcha++ if ($v =~ s/^\s*(.+str)(\.?|aße)\s*$/$1asse/i); 					# am ende [s|S]tr[.|] => [s|S]trasse [s|S]traße => [s|S]trasse
     $gotcha++ if ($v =~ s/^\s*(.{4,}str|.{4,}(\s|-)Str)(\.?|aße)((\s+|,|;|\/).+)\s*$/$1asse$4/);	# das ganze in der mitte, vor [ |,|;|/]
     $gotcha++ if ($v =~ s/^\s*(.{4,}(\s+|-))strasse(.*)\s*/$1Strasse$3/);				# Großschreibung nach Leerstelle oder '-'
     $gotcha++ if ($v =~ s/ß/ss/ );									# Generelle Ersetzung von ß durch ss
     $gotcha++ if ($v =~ /strasse/i && ($v =~ s/\s{2,}/ /g || $v =~ s/-{2,}/-/g) );			# Mehrfache Leerstelle oder '-' durch einzelne ersetzen
     if ($gotcha) {
       $resultstr .= sprintf " - key '%s' modifying value '%s' to '%s'", $k, $work->{"tags"}->{$k}, $v;

       if (defined($work->{"tags"}->{"is_in"}) && $work->{"tags"}->{"is_in"} =~ /Deutschland|Germany|DE|sterreich|Austria|AT/) {

1. fs $obj->log(ref($obj), $work->{"id"}, $resultstr, "NOTICE not modifying - has tag 'is_in=".$work->{"tags"}->{"is_in"}."' ");

       } elsif (defined($work->{"tags"}->{"addr:country"}) && $work->{"tags"}->{"addr:country"} =~ /DE|AT/) {

1. fs $obj->log(ref($obj), $work->{"id"}, $resultstr, "NOTICE not modifying - has tag 'addr:country=".$work->{"tags"}->{"addr:country"}."' ");

       } else {
         $clone=$work->clone() unless defined($clone);
         $clone->{"tags"}->{$k} = $v;
         $work=$clone;
       }
     }
     if ( $v =~ /^\s*stra(ss|ß)e\s*$/i ) {
       $clone=$work->clone() unless defined($clone);
       delete $clone->{"tags"}->{$k};
       $resultstr .= sprintf " - deleting meaningless tag '%s=%s'", $k, $v,
       $work=$clone;
     }
   }
 }